On Wednesday 28 April 2004, Leopold Toetsch wrote: > Jens Rieks <[EMAIL PROTECTED]> wrote: > > Leo, what is the status of the pending changes WRT library paths? > > Mainly a missing portable way to find the library. *But* searching in > C<runtime/parrot/include> is working for current platforms, so we can > do the same for /library. We just need to abstract it a bit and put a > wrapper into the platforms file(s). It can get refined always and > gradually. > > > Can I help somehow? > > We would need a more general version of dynext.c:get_path() [1] that just > tests for existence of the file: > > STRING * Parrot_get_runtime_path(interp, file_type, filename) > > or some such. File type is one of [dynamic/shared extension, include_file, > library] currently. I had no luck with writing this function, mainly because I do not know much about parrots internals, especially when it comes to filename/path handling and testing if a particular file exists (without doing "bruteforce" opens).
I can implement it in bytecode, but then this bytecode needs to be linked with parrot. The advantage is of course that it is then very easy to extend the mechanism. > The searchorder should be configurable finally, for now current directory > and runtime/parrot/* should do it. Is runtime/parrot/include and runtime/parrot okay for the moment? Code that currently uses load_bytecode "library/foo.imc" will continue to work, even if we move the library directory to runtime/parrot/library. > [1] and imcc/imcc.l:include_file() How can I regenerate the corresponding C files? I looks like there's no Makefile rule for it, intentionally not? jens
